Energy Digest

A deep learning-based framework is proposed to detect electrical faults and power quality disturbances in aerospace power systems, using a 400 Hz system and achieving high accuracy with a compact ResNet model. The model achieves 96.94% software test accuracy after deployment on an Xilinx Zynq UltraScale Plus MPSoC. The results demonstrate the feasibility of edge AI for aircraft electrical health monitoring.

QZ-IPCA recovers the noisy-variable partition, noise variances, number of linear relations, and regression coefficients simultaneously from raw data without prior assumptions. It outperforms ordinary least squares even when given the best partition and succeeds in rank identification precisely where standard total least squares falls. The algorithm achieves an error rate of below 6.4% across all possible exhaustive noise configurations in a five-variable benchmark network.

The US interconnection queues contain approximately 2,061 GW of capacity and have only 13% of queued projects operational since 2000-2020. The system is vulnerable to self-reinforcing project withdrawal cascades due to restudy and cost reallocation, which can amplify by up to 8.7x under certain conditions. Capping per-neighbor cost reallocation reduces cascade magnitude, highlighting the need for a more robust design in the interconnection queue's cost allocation mechanism.

AI agents can't guarantee network-wide outcomes due to operational networks spanning multiple devices and administrative domains with distinct authority scopes. A trusted assurance layer is needed to collect current observations and verify whether they support an operator's intended outcome. EvidenceNet, a runtime assurance layer, addresses the completion admission problem by verifying observation content and ensuring that evidence comes from required scopes.

AI data centers may need to shift LLM inference workloads spatially to maintain service rates during power supply shortages, but existing methods assume any site with sufficient resources can handle the shift, which is often infeasible. A new framework called model commitment (MC) solves this by jointly scheduling model deployment and request routing under power constraints and electricity-price signals. MC enables AI data center operators to achieve 100% service rates under time-varying grid conditions and reduce total operating costs by 29.0%.

Existing PLA schemes in non-terrestrial networks often lack joint authentication-transmission design and prioritize tag privacy, but ignore these issues under passive, location-aware eavesdroppers. A proposed method, SAFA-MZ, uses group-level authentication tags to maximize secrecy spectral efficiency while ensuring authentication reliability, power limits, and coverage constraints. The method improves average secrecy spectral efficiency by up to 135% over single-connect transmission in simulated scenarios.

Fast charging stations with V2G (vehicle-to-grid) technology can optimize EV adoption and improve the economic viability of transportation electrification by mitigating power grid vulnerabilities during peak traffic hours. A new framework evaluates the economic benefits of V2G-enabled FCSs across both transportation and distribution networks, achieving superior expected social welfare through macroscopic vehicle-to-grid aggregation. This approach simplifies complex calculations and ensures calculation speed while guaranteeing calculation accuracy.

Predicting the lifespan of residential battery energy storage systems involves bridging the gap between laboratory aging models and field-level accuracy with a probabilistic degradation framework. The framework, which approximates cell-level heterogeneity using two bounding stress scenarios, achieves mean absolute error of 0.4% for state-of-health predictions and aligns with system-level capacity measurements. The uncertainty is linked to mismatches in test conditions and training data variability, leading to recommendations for more accurate aging study design.

A hydro-gen colocated renewable-powered desalination plant can dispatch water, electricity, and green hydrogen in an optimal manner that maximizes profits, relying on thermal and reverse osmosis desalination at low renewable output and increasing RO desalination and hydrogen production at high outputs. The optimal schedule is characterized by closed-form functions of renewable generation, balancing energy internally to allocate resources based on the relative marginal values of hydrogen and RO water. This configuration achieves a daily profit of $75.945 k$, exceeding the best configuration without hydrogen by 15.1%.

The Arctic regions heavily rely on fossil fuels due to their high dependence on energy generation. State-of-the-art energy solutions like wind, solar, hydrogen, batteries, and thermal storage show potential to reduce fossil fuel dependence and meet heating demands in the region. However, technological and social barriers need to be addressed through advancements in planning tools and equipment, governmental support, subsidies, and legal frameworks to enable widespread adoption of low-carbon energy systems.